"""Settings and configuration for the MCP server."""
import os
# ============================================================================
# AI Configuration
# ============================================================================
# To use AI-powered classification, configure one of the following:
# Option 1: OpenAI API
# Set via environment variable: OPENAI_API_KEY='sk-...'
# This will use GPT-3.5-turbo for classification
OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY")
# Option 2: Custom external LLM endpoint
# Set via environment variables: EXTERNAL_LLM_URL and EXTERNAL_LLM_API_KEY
# The endpoint should accept JSON: {"filename", "file_type", "content_preview"}
# And return JSON: {"category", "confidence", "subcategory", "suggested_name"}
EXTERNAL_LLM_URL = os.environ.get("EXTERNAL_LLM_URL")
EXTERNAL_LLM_API_KEY = os.environ.get("EXTERNAL_LLM_API_KEY")
# Option 3: Interactive mode (for MCP Inspector)
# When INTERACTIVE_MODE is True and AI is unavailable, the system will
# prompt the user via stdin for classification decisions.
INTERACTIVE_MODE = os.environ.get("INTERACTIVE_MODE", "false").lower() == "true"
# ============================================================================
# File Processing Settings
# ============================================================================
MAX_FILE_SIZE_MB = 10
# Maximum number of files to analyze per folder
MAX_FILES_TO_ANALYZE = 1000
# File extensions to consider as text
TEXT_FILE_EXTENSIONS = {
".txt",
".md",
".py",
".js",
".ts",
".json",
".yaml",
".yml",
".xml",
".html",
".css",
".sql",
".java",
".cpp",
".c",
".h",
".sh",
".bash",
".log",
".csv",
".ini",
".conf",
".config",
".toml",
".properties",
".vue",
".jsx",
".tsx",
".scala",
".rb",
".php",
".go",
".rs",
}
# Category detection keywords
CATEGORY_KEYWORDS = {
"Documents": [
"document",
"doc",
"pdf",
"word",
"text",
"report",
"essay",
"article",
"paper",
"memo",
"letter",
"contract",
"agreement",
],
"Images": [
"image",
"photo",
"picture",
"img",
"graphic",
"design",
"icon",
"screenshot",
],
"Videos": ["video", "movie", "film", "clip", "recording"],
"Audio": ["audio", "music", "sound", "song", "podcast"],
"Code": [
"code",
"script",
"program",
"source",
"python",
"java",
"javascript",
],
"Data": ["data", "dataset", "database", "csv", "sql", "excel", "sheet"],
"Configuration": [
"config",
"settings",
"setup",
"configuration",
],
}
# Maximum folder name length
MAX_FOLDER_NAME_LENGTH = 200
# Enable detailed logging
DEBUG = False
# ============================================================================
# Configuration Setup Guide
# ============================================================================
#
# To get started with AI classification:
#
# 1. Using OpenAI (recommended for quick start):
# powershell> $env:OPENAI_API_KEY = 'sk-your-key-here'
# powershell> python main.py /path/to/folder
#
# 2. Using custom external LLM:
# powershell> $env:EXTERNAL_LLM_URL = 'https://your-api.example.com/classify'
# powershell> $env:EXTERNAL_LLM_API_KEY = 'your-api-key'
# powershell> python main.py /path/to/folder
#
# 3. Interactive mode (ask user for classification):
# powershell> $env:INTERACTIVE_MODE = 'true'
# powershell> python main.py /path/to/folder
#
# Without any AI configuration, files will be categorized as 'Uncategorized'.
